Modern electronics have benefited from the ability to fabricate devices on a smaller and smaller scale. As the ability to shrink devices has improved, so has their performance. Unfortunately, this improvement in performance is accompanied by an increase in power as well as power density in devices. In order to maintain the reliability of these devices, the industry must find new methods to remove this heat efficiently.
By definition, heat sinking means that one attaches a cooling device to a heat-generating component and thereby removes the heat to some cooling medium, such as air or water. Unfortunately, one of the major problems in joining two devices to transfer heat is that a thermal interface is created at the junction. This thermal interface is characterized by a thermal contact impedance. Thermal contact impedance is a function of contact pressure and the absence or presence of material filling small gaps or surface variations in the interface.
As the power density of electronic devices increases, heat transfer from the heat generating devices to the surrounding environment becomes more and more critical to the proper operation of the devices. Many current electronic devices incorporate heat sink fins to dissipate heat to the surrounding air moving over the fins. These heat sinks are thermally connected to the electronic devices by a variety of techniques. Some devices use a thermally conductive paste in an attempt to lower the contact resistance. Others may use solder between the two elements both for mechanical strength and thermal conductance. However, these two solutions require additional cost and process steps that would not be necessary except for presence of the contact resistance, and also only work for small gap sizes on the order of a few mils.
The heat-sinking problem is particularly difficult in devices such as multi-chip modules (“MCMs”) where multiple components need to have topside cooling into a single cold plate or heat sink. The various components within the multi-chip module may not be of equal thickness, creating a non-coplanar surface that often must be contacted to a single planar surface of the cold plate or heat sink. Engineers have developed a variety of approaches to solving the non-coplanar surface problem, such as, gap fillers comprising thick thermal pads capable of absorbing 10 to 20 mils of stack up differences. However, the thickness and composition of these thermal pads often results in a relatively high thermal resistance making them suitable only for low power devices. Others have used pistons with springs attached to them attached to a plurality of small cold plates or heat sinks to account for the irregularity of the stack up. However, this can become an expensive solution to the problem. Still others have used an array of small cold plates connected together by flexible tubing allowing some flexibility between the plates to account for the variations in height of the components. However, once again, this solution may become too expensive for many products.
Other solutions include the use of thermal grease or phase change materials, such as paraffin, to fill in small gaps, such as the microscopic roughness between two surfaces. However, thermal grease and phase change materials are unable to fill larger gaps such as those present in multi-chip modules.